An Ex Vivo Explant Model for Studying Glial Interactions in the Mouse Retina

The light sensitive retina, an extension of the central nervous system (CNS) located at the back of the eye, is essential for sight but vulnerable to both acute injury and chronic disease. As with other CNS regions, neurons in the adult retina are not replaced when lost, and the retinal ganglion cells (RGCs) that aggregate and relay visual information to the brain are particularly vulnerable to dysfunction and death in glaucoma, resulting in irreversible vision loss^1,2. Glaucoma is a leading cause of blindness worldwide³, yet despite the devastating impact on affected individuals and the overall costs to society, much remains unknown about how the early stages of the disease contribute to RGC loss⁴. The role of glia is a major area of investigation in glaucoma pathophysiology, as these non-neuronal support cells are essential for RGC survival⁵ but undergo phenotypic changes in disease that may diminish beneficial behavior⁶ or even drive the adoption of deleterious phenotypes⁷. Although the term glia encompasses a range of specialized cell types throughout the CNS⁸, these can be broadly divided into two categories - those that share a developmental lineage with neurons and provide trophic, energetic, and structural support⁹, and those with a myeloid origin, which perform specialized immune surveillance and response tasks¹⁰. Representatives of both categories - astrocytes and Müller cells in the former, microglia in the latter - populate the retina⁵ and undergo major changes in glaucoma that raise significant questions about their role in disease progression and RGC survival¹¹.

Efforts to answer these questions are hindered in part by intrinsic characteristics of retinal glia that present challenges to both in vivo and in vitro investigation. Unlike neurons, they are relatively silent electrically, making approaches such as ERG that enable functional assessment of RGCs and photoreceptors in vivo unsuitable for investigating changes in glial function and behavior. And while inducible¹² and spontaneous¹³ models of glaucoma are available, much remains unknown about changes in glia at the earliest time points in the disease, which may precede detectable RGC loss, a problem compounded by the variable penetrance of disease state in many of these models. Furthermore, evidence from both clinical and experimental glaucoma suggests contributions from peripheral immune cells that can be difficult to disambiguate from those of microglia, despite indicators that they may act in different 'directions' to influence disease progression^14,15.

Challenges also abound in studying retinal glia in vitro. Although isolation and culture of astrocytes^16,17 and microglia¹⁸ from the brain are well established, recent work highlights extensive glial heterogeneity between CNS regions, especially in astrocytes¹⁹, and phenotypes present in one region may not be present in glia from another, such as the retina^20,21. However, directly isolating retinal astrocytes and microglia for study is particularly challenging, as both cell types are relatively sparse - each making up less than 1% of the estimated 6.5 million cells in the mouse retina^22,23,24. Furthermore, unlike neurons, glia are highly plastic and rapidly adapt to dramatic changes in their surroundings^16,18,25; as a result, behaviors observed in these cells in vitro may represent specific adaptations to their new environment rather than phenotypic patterns that would be typically seen in health or disease. Given the limitations of both in vivo experimentation and primary culture of retinal glia, we have sought an intermediate approach - retinal explants - in which glia, RGCs, and other elements of the retina are preserved in situ in an ex vivo context. Relative to in vivo models, this approach offers an abbreviated experimental time course²⁶, enables direct experimental manipulation of retinal glia²⁷, and avoids the potentially confounding influence of peripheral immune cell infiltration¹⁴. Conversely, unlike primary cell culture, there is minimal disruption of the extracellular environment, and glia remain intact and morphologically recognizable, obviating the challenges associated with essentially regrowing and identifying these cells after enzymatic and mechanical disruption^16,25.

Relative to previously described retinal explant methodologies, this approach emphasizes a focus on technical reliability, reproducibility, and maximizing the 'user-friendliness' of the approach to improve accessibility^26,28. In personal communications with other researchers, we found that the technical challenges associated with handling the live retina present a major hurdle to many looking to utilize explants, whereas maintenance of the explanted retina in culture was relatively straightforward for groups with appropriate cell culture facilities. Therefore, this protocol includes a number of innovations intended to reduce the learning curve associated with retina isolation and allow researchers to more rapidly begin collecting experimental data. Finally, although we have placed special emphasis on the potential of this explant model for investigating the behavior of retinal glia and characterize it primarily with immunofluorescence microscopy, other retinal cell types and structures are largely conserved as well, and the explanted retinas remain amenable to a wide range of additional investigatory techniques.

All procedures involving animals were approved by the Institutional Animal Care and Use Committee (IACUC) at Schepens Eye Research Institute (Protocol # 2022N000030, approved 3/4/2025). Animals were handled in accordance with the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research.

NOTE: Figure 1 shows the key steps in the isolation of live retina from the enucleated mouse eye.

Figure 1: A diagrammatic representation of key steps in the isolation of live retina from the enucleated mouse eye. The corresponding step numbers of the key steps are provided in circles. Please click here to view a larger version of this figure.

1. Preparation

1. Disinfect a 1 mL pipet and a pair of angled forceps and place them in the biosafety cabinet. These will be needed in section 3 of the protocol.
2. Modify a transfer pipet by cutting perpendicularly across the shaft approximately 2 cm below the reservoir, where the bore is widest, to shorten the pipet and widen the opening. Ensure a clean cut on which the retina will not snag.
   NOTE: This tool can be kept for reuse if clean and rinsed with sterile water between uses.
3. Prepare 50 ml of explant media aseptically in a biosafety cabinet by supplementing 47.5 mL of Neurobasal media with 500 µL of N-2 supplement, 1 mL of B-27 supplement, 500 µL of Glutamax, and 500 µL of penicillin-streptomycin mix (final concentration of 100 U/mL of penicillin and 100 mg/mL of streptomycin).
   NOTE: Prepare fresh explant media for each series of isolations; excess media can be kept at 4 °C for up to 1 week and used for media changes if needed. After opening, replace unused Neurobasal medium after 1 month. N-2 and B-27 should be kept frozen as single-use aliquots, avoiding freeze-thaw cycles.
4. In a biosafety cabinet, aseptically prepare culture plate(s) by adding an insert into each well to be used, handling via the 'prongs' that emerge from the insert rim to avoid touching surfaces that will contact culture media.
   1. Add 1 mL of explant media to the insert, followed by an additional 3 mL added to the well itself. Add 2-3 mL of sterile water to any unused wells to control evaporation.
   2. Transfer plate(s) to the incubator at 37 °C with 5% CO₂ to equilibrate approximately 30 min before tissue isolation.
5. Prepare filter squares to create a substrate for tissue stabilization and transfer. Extract the filter from a bottle-top filtration kit by cutting around the periphery of the material with a razor or scalpel. Take care to avoid injury or damaging the filter through folding or tearing.
   NOTE: Removing the clear plastic reservoir using a sturdy razor to cut the more malleable opaque plastic where the threaded adapter meets the reservoir simplifies the extraction process, but additional caution should be used to prevent injury.
   1. After the filter has been detached by cutting, use forceps to extract it. Note the orientation of the filter during and after extraction; the matte side (which faces the reservoir) should be upright for mounting the retina in step 2.16 for transfer.
   2. Cut the filter into small squares approximately 7 mm × 7 mm; each retinal explant will require one square. Pre-soak the squares in sterile water in a 100 mm Petri dish for at least 20 min.
      NOTE: Filter squares can remain immersed in sterile water for up to 1 week prior to use. Due to the tendency of the filter squares to develop high levels of static charge prior to immersion, prepare only as many squares as needed. The remaining dry filter material can be kept indefinitely in a sterile 100 mm Petri dish for later use.

2. Isolation and mounting

1. Euthanize the mouse via an approved method before confirming with a secondary approach; this study used CO2 asphyxiation followed by cervical dislocation. Enucleate both eyes with curved, blunt-tipped forceps and place them in sterile PBS.
2. Prepare the workspace by filling a 100 mm Petri dish with sterile PBS and placing it on the stage of the binocular dissection scope.
   1. Cut a 1 cm wide strip of lab wipe and submerge it in the dish to act as a stabilizing substrate.
   2. Transfer one eye to the dissection dish, leave the other in PBS, and place on ice or in a refrigerator at 4 °C.
      NOTE: Due to the retina's sensitivity to prolonged postmortem interval, euthanizing one mouse at a time is recommended.
3. Position and immobilize the eye. Use forceps to maneuver the eye onto the submerged lab wipe, which will aid stabilization.
   1. Identify an appropriate holding point, grasp it with angled forceps, and gently roll the eye so that it is held from the side or, ideally, from below. Ensure that the anterior-posterior axis (i.e., from the cornea to the optic nerve) is horizontal. This orientation improves visibility and minimizes pressure on the eye during dissection.
      NOTE: If possible, grasp extraorbital tissue to avoid puncturing or squeezing the eye, favoring muscle or conjunctiva over optic nerve or fat.
4. While maintaining a hold with the angled forceps, ensure the eye is fully submerged and securely immobilized. Make an incision with the tip of the #11 scalpel parallel and approximately 0.5 mm posterior to the limbus, where the cornea transitions to the sclera.
   NOTE: This incision should be just large enough to fit the tip of one blade of the spring scissors. As the sharpness of the scalpel tip is essential to piercing the globe cleanly, change the blade after every two retinas.
   1. Insert one blade of the spring scissors inside the globe and begin cutting circumlimbally around the eye, gently repositioning the eye as needed using forceps.
      NOTE: This cut should be as straight and parallel to the limbus as possible; repositioning the eye may require putting down the scissors and using one pair of forceps with each hand. Once exposed, keep the retina submerged throughout the dissection.
5. Once the cut has circumscribed the eye, carefully remove the anterior segment and lens, then rotate the eyecup to face upward to ease visual inspection and vitreous removal. If an extended piece of optic nerve remains attached, trim it to a shorter (1-2 mm) length to facilitate this.
6. Inspect the retina for visible damage and the vitreal chamber for debris, such as pigmented cells from the RPE or choroid that may have entered during the removal of the anterior segment. Use angled forceps to immobilize the eyecup in this position.
   NOTE: For optimal results, debris in the vitreous chamber must be removed, and residual vitreous and adherent remnants of the ciliary body at the retinal periphery must be minimized. This can be a challenging process, and additional notes can be found in the 'Recognizing and troubleshooting mechanical injury' and 'Metabolic injury' sections of the Representative Results.
   1. Use the modified transfer pipet to flush the vitreous chamber with PBS to remove small particulate debris, avoiding air bubbles by keeping the pipet tip below the surface of the liquid and maintaining an adequate amount of PBS in the reservoir bulb.
   2. Remove larger debris with a fine watercolor brush, minimizing contact with the retinal surface. If necessary, remove more persistent debris with fine-tipped forceps; however, avoid direct contact of metal tool tips with the retina to prevent damage.
7. After clearing visible debris, flush the vitreous chamber with PBS from the transfer pipet repeatedly (3-5 times), as described above. This will remove loosely adherent vitreous that may remain after debris removal. Use the fine brush to gently probe the chamber, particularly near the periphery if elements of the ciliary body remain. While the vitreous is optically transparent, it will act as a drag on brush fibers and can be detected in this way.
   NOTE: A thin residual layer of vitreous on the retina does not appear to negatively impact isolation or keeping the retina ex vivo, but the presence of significant amounts at the retinal periphery, often associated with residual structures from the ciliary body, can cause retinal folding and sample loss if not removed.
   1. If significant pockets of vitreous remain, use the brush to remove it by gently sweeping outward towards the periphery, ensuring that if the fibers of the brush contact the retina, they do so from a shallow angle and trail the brush, pulling rather than pushing vitreous from the chamber.
      NOTE: The removal of vitreous requires particularly careful handling to avoid retinal damage that may not be visible macroscopically; advice on recognizing and avoiding this damage is found in the 'Recognizing and troubleshooting mechanical injury' and 'Metabolic injury' sections of the Representative Results.
8. Once the vitreous chamber has been cleared of debris and only minimal vitreous remains, gently separate the outer retina from the eye cup while continuing to stabilize the sample with angled forceps. Avoid damaging the optic nerve head and leave this structure intact to anchor the retina.
   1. Take a pair of forceps, keeping them in closed position with tips touching, and gently insert them between retina and choroid, using a pre-existing gap from prior handling if possible. Use the flat arms of the forceps, rather than the tips, to enlarge the pocket between the retina and choroid until these structures can be completely separated.
      NOTE: Emphasize lateral motions, rather than pushing the tips, to expand the gap between the retina and choroid. Gently twisting the forceps during these motions can improve separation while reducing sheer stress from friction between forceps and tissue.
9. After separating the outer retina from the choroid, continue stabilizing the sample while using a second pair of forceps to pull the eyecup - sclera, choroid, etc. - downward. If the retina has been successfully separated, it should remain in place, tethered at the optic nerve head, while the eye cup can be held by a single pair of forceps. If the retina presses down with the eyecup, continue gently probing and separating points of connection other than the optic nerve head.
10. With the retina anchored at the optic nerve head, continue immobilizing the tissue and use the second pair of forceps to bunch up the eye cup below the optic nerve head.
    1. Check to ensure that none of the retinal periphery is stuck in a folded-over position due to excess vitreous in the angle between the retina and any residual ciliary body. If necessary, flush the chamber with PBS using the transfer pipet to remove any additional debris, and gently use the brush to uncurl the tissue and eliminate excess vitreous, while taking care not to tug the tissue excessively.
11. With the retina exposed from both sides, use the spring scissors to make a series of relieving cuts, spaced approximately 90° apart, extending in a straight line from the periphery towards the optic nerve head, stopping approximately 1 mm from the center. This minimizes the transection of retinal ganglion cell axons in the retina, improving the quality of cultured explants.
12. While holding the submerged tissue in place, gently remove the lab wipe using forceps to pull it away from the tissue laterally before removing it from the dish. Avoid contact with the retina to prevent adhesion to the wipe and possible sample loss.
13. Next, continue to hold the tissue and use the spring scissors to sever the optic nerve directly beneath the retina, then remove the remainder of the eyecup from the dish.
14. Once the retina is isolated from the rest of the eye, fill a 35 mm Petri dish with PBS and use forceps to place a filter square at the bottom with the rougher matte side facing up. Take care to avoid creasing the filter, as it may impede the use of the filter as a substrate for moving the retina.
15. Gently aspirate the retina with the transfer pipet and carefully transfer it to the 35 mm dish. Maneuver the retina with the brush to ensure the inner surface of the retina is upright, and the tissue rests directly above the filter square.
16. Slowly aspirate PBS from the dish with the transfer pipet to lower the retina onto the filter. Pause, if necessary, to readjust the tissue before continuing until the retina has settled onto the filter.
    NOTE: Rapid aspiration can drive sudden lateral motion of the retina, resulting in damage or tissue loss. The entirety of the retina must rest on the square to avoid damage; however, if the retina settles in a problematic position, PBS can be added back to the dish until the retina is resuspended, and mounting can be re-attempted.
17. Once the retina has settled onto the filter, use the brush to unroll any peripheral retina that may have folded over. Carefully balance the level of PBS so that the retina does not easily move from the filter while ensuring that enough remains so that the sample does not dry out and the brush can move smoothly without damaging the inner retina.
18. Once again inspect the retina for debris. To avoid damage, clean the retina without direct handling using the transfer pipet to drip PBS onto the retina from roughly 1 cm above the tissue. This should flush away any remaining debris, but if too much PBS is added to the dish, the retina may float back up. If this occurs, perform aspiration again to remount the tissue.
    NOTE: 'Sham' samples can be generated by transferring retinas to a 24-well plate for fixation, blocking/permeabilization, and antibody staining as described elsewhere²¹. These sham controls are useful for troubleshooting mechanical injury to the retina that may occur during isolation.

3. Transfer to culture media and upkeep

1. Close the lid of the 35 mm dish for transport to the biosafety cabinet, keeping the dish level to avoid dislodging the retina from the filter. If this occurs, remount as above.
2. Place the closed 35 mm dish containing the retina in the biosafety cabinet, taking care not to touch any equipment or surfaces within the cabinet before disinfecting or changing gloves.
3. Decontaminate gloves with ethanol and/or replace them, then move the 6-well plate pre-loaded with explant media from the incubator to the biosafety cabinet. Perform steps 3.4 and 3.5 within the biosafety cabinet, using the tools placed there during step 1.1.
4. Remove the lid of the 35 mm dish containing the retina and use angled forceps to carefully lift the filter square without touching the retina directly. Remove the lid of the 6-well plate and position the filter over the center of the insert of an appropriate well, then slowly lower it into the media. The retina will detach from the filter during this step and float just below the air-liquid interface.
5. Once the retina and filter have separated, slowly move the filter away from the retina and remove it from the media. Using the 1 mL pipet, remove 500 µL of media from the insert to ensure the retina lies flat by trapping it between the floor of the insert and the air-liquid interface.
6. Replace the lid on the plate and return it to the incubator at 37 °C with 5% CO₂, taking care to keep the retina near the center of the well. Up to 6 retinas can be kept in a single plate.
   NOTE: If keeping explants for 24 h or less, they can be left in the incubator until the end of the experiment. If kept for longer durations, 50% media changes after 1 day in vitro and on alternating days thereafter are recommended.

4. Fixation for immunostaining (optional)

1. If performing immunostaining, refer to the following steps for fixation; while immunostaining can be performed using parameters previously detailed for flat-mounted retinas²¹, fixation of the explanted retina requires specific considerations.
2. Remove cell culture inserts with explants and media from the 6-well culture plate and place them on a non-porous surface (such as the plastic lid of the culture plate) that can be disposed of properly after fixation, as the inserts are slightly porous and small amounts of fixative will pass through to this underlying surface.
3. Gently remove culture media from inserts, then add 1 mL of 4% paraformaldehyde (PFA) to each. Let retinas fix for 15 min in a fume hood, then remove PFA and dispose of it according to institutional policy.
   CAUTION: Paraformaldehyde is hazardous and a precursor of formaldehyde, a known carcinogen. Exercise caution and perform fixation in a fume hood with suitable precautions.
4. Wash retinas 3 times (for 5 min each) with PBS to remove PFA, then use a modified transfer pipet to transfer the retinas in PBS to individual wells in a 24-well plate for blocking, permeabilization, and immunostaining as described elsewhere²¹. Dispose of PFA-contaminated plastics (pipet tips, inserts, and the non-porous surface used during fixation) according to institutional policy.
   NOTE: The transfer pipet can be kept to transfer fixed and stained retinas to a slide for mounting and imaging. Although it should not have had direct contact with PFA, using different transfer pipets for handling live and fixed tissue is recommended.

At the macroscopic scale, explanted retinas remain essentially unchanged at both 1- and 3-day time points, although by the latter time they become relatively fragile, necessitating careful handling prior to fixation or other experimental endpoints. Retinas should be relatively flat, without folding (which can result in localized disruption to the diffusion of oxygen and nutrients), and should be suspended in the media between the air-liquid interface and the cell culture insert. By day 3, a faint streak may form on the bottom of the insert beneath the suspended tissue; inspection with brightfield phase contrast microscopy suggests a moderate amount of shedding of cellular debris, likely the outer segments of degenerating photoreceptors, particularly between the 24-h and 72-h time points. This is unsurprising, given the fragility of photoreceptors and their reliance on an RPE substrate that is largely lost during retinal isolation. However, if the media itself appears cloudy, the possibility of microbial contamination should be investigated. Despite handling the retina outside of a biosafety cabinet during the initial isolation process, we were unable to detect microbial contamination in any of our samples, suggesting that the use of sterile buffers, the inclusion of penicillin (100 U/mL), and streptomycin (100 µg/mL) in culture media, and careful aseptic technique in the tissue culture environment were sufficient to avoid detectable contamination within the studied time frame.

Microscopic assessment via immunofluorescence
In the retina, both astrocytes and microglia (along with ganglion cells) typically orient their cell bodies parallel to the retinal surface, making them well-suited for assessment with immunofluorescence microscopy. To probe large-scale changes in retinal glia, we compared retinas that had been isolated and kept as explants for 3 days with naïve 'sham' explants, which underwent identical isolation procedures until step 2.18 of the protocol, at which point they were fixed for staining (see steps 4.1-4.4) rather than kept in vitro. While individual astrocytes and microglia undergo phenotypic changes (reactivity in astrocytes and activation in microglia, see below) as a result of changes in the explanted retina, large scale changes in these populations are also visible after 3 days in vitro (Figure 2), with microglia showing signs of migrating from their typically non-overlapping domains (Figure 2A,D) and astrocytes reducing the closeness of their association with the superficial retinal vasculature (Figure 2B,E).

Although investigating RGCs is not the primary focus of this protocol, we assessed their density alongside astrocyte and microglia response at 24-h and 72-h ex vivo (Figure 3A-L) and quantified RGC survival at the 24-h time point (Figure 3M), as it is a valuable metric to consider when evaluating changes to glial phenotypes. Furthermore, while probing astrocytes and microglia in greater detail, changes were also detected in Müller cells (Figure 2B,E) - specialized radial glia of the retina - and hyalocytes (Figure 4), a class of border-associated macrophage found on the vitreal side of the retinovitreal interface. For both qualitative and quantitative imaging, samples were fixed and stained with commonly used markers of astrocytes (GFAP), microglia (Iba1), and RGCs (Brn3a). This study additionally utilized the homeostatic microglial marker TMEM119 to probe changes in microglia (Figure 4C,G) and the hyalocyte marker CD206 (Figure 4B,F) to differentiate these two cell types, which both express Iba1 (Figure 4A,E).

Retinal ganglion cell survival
At the 24-h mark, RGC density showed a modest but marked decline over sham explants (Figure 3M). For both 24-h and sham retinas (n = 4 for each condition), 9 mid-periphery regions of interest per retina (36 total) were imaged, and Brn3a+ cells across each imaged region were manually counted using the cell counter module in ImageJ. This approach yielded a density of 3659 RGCs per mm2 (SD +/- 490) in sham retinas, which matches prior reports for adult mice of this strain (the widely utilized C57BL/6)24. Counts on retinas after 24 h in vitro indicated a density of 2464 RGCs per mm2 (SD +/- 907), a decline of 32.7% relative to shams. While a 72-h time point was also investigated, survival was difficult to accurately quantify at this time point due to several factors, including high regional variability within individual retinas and a significant increase in background in Brn3a staining on these samples. Given the clarity of staining for other markers used on samples at this time point - including GFAP, Iba1, CD206, and TMEM119 - we suspect this stems from the use of a mouse-derived anti-Brn3a antibody and would recommend the use of an antibody raised in a different host species for investigators seeking to more precisely quantify RGC survival on explanted tissue.

Microglia and hyalocytes
To qualitatively assess changes in microglia, we utilized the myeloid cell marker Iba1, which is strongly expressed by microglia in the retina and elsewhere. Microglia engage in dramatic morphological changes when activated by stress or injury, which can be assessed even without additional markers, and we observed a moderate level of morphological change after 24 h in vitro (Figure 3F), with microglial processes shortening and decreasing in number. By the 72-h time point, this transformation had progressed still further, with many microglia adopting a compact amoeboid morphology with few or no visible processes (Figure 3J). A change in microglial distribution was also noted, as microglia transitioned from the regular, non-overlapping spacing seen in the sham retina (Figure 2A) to regional aggregation seen at the 72-h time point (Figure 2D). Additionally, expression of the homeostatic microglial marker TMEM119 declined dramatically from clear and ubiquitous levels observed in sham retinas (Figure 4C) to a near total absence of immunoreactivity after 3 days in vitro (Figure 4G). Taken together, this evidence points to a major shift towards an activated phenotype in retinal microglia resulting from changes in the retina ex vivo.

While developing this protocol, we also noted that we were capturing hyalocytes, a population of border-associated macrophages (BAMs) found at the vitreoretinal surface29. These cells, which like microglia express Iba1, tend to be more amoeboid than non-activated microglia and do not express TMEM119; however, these distinctions are typically lost by the 72-h mark in vitro as microglia become more amoeboid and downregulate TMEM119 expression (Figure 4). However, under these conditions, hyalocytes (which are much less dense than microglia, with published estimates reporting29 approximately 10 per square mm) can be distinguished by their expression of CD206, which is conserved across investigated time points (Figure 4B,F). The presence of hyalocytes in isolated samples is highly dependent on the extent of vitreal removal during sample preparation, and aggressive removal of the vitreous eliminates these cells entirely in the mid-periphery.

Astrocytes and Müller Cells
In contrast, astrocytes show relatively modest signatures of reactivity and, at the 24-h and 72-h time points, maintain their characteristic tiling of the inner retinal surface, contacts with the retinal vasculature, and only moderate indication of hypertrophy and structural changes (Figure 2B,E; Figure 3C,G,K). Conversely, although Müller cells maintain low-to-undetectable GFAP expression at day one (similar to shams), by day 3 Müller cell expression of GFAP is increasingly detectable. This is particularly evident near the edges of the tissue - both at the retina's periphery and along the relieving cuts made during retinal dissection to enable the tissue to lie flat in culture (Figure 2E). Groups that have investigated explants in culture for longer durations have described a dramatic increase in markers of astrocyte and Müller cell reactivity after the 3-day time point30. However, the emergence of an apparent delay between the response of RGCs and microglia and that of astrocytes within this relatively brief time frame is itself striking and may provide hints to the phenomena governing changes in these explanted retinas.

The relatively homeostatic appearance of astrocytes and Müller cells also makes them excellent indicators of physical damage sustained by live tissue. In areas where the tissue has been either cut deliberately (relieving cuts) or damaged inadvertently, astrocytes and Müller cells strongly upregulate GFAP, highlighting areas of mechanical damage to the tissue that might otherwise be masked (Figure 4J). Conversely, due to their close association with the inner limiting membrane (ILM), overly aggressive vitreous removal that damages the ILM can result in major astrocytic damage within a region (Figure 4K), potentially including the complete loss of astrocytes therein. Such events were observed to be associated with poor experimental outcomes.

Recognizing and troubleshooting mechanical injury
As a thin sheet of neural tissue, the retina is highly vulnerable to both mechanical injury and metabolic disruption, and most difficulties with successfully isolating the retina and keeping it in culture as an explant stem from these vulnerabilities. Mechanical injury occurs primarily during isolation and is often straightforward to identify but may require significant practice to minimize. Some forms of damage - such as tearing of the retina due to adhesion of the periphery to the ciliary body - are readily apparent, enabling the exclusion of affected samples. More subtle injuries are also possible and can nonetheless impact experimental outcomes, but fortunately these can be detected with immunofluorescence microscopy as described below.

The chief risk for damage during retina isolation is during the removal of excess vitreous, which can be a challenge even for those with experience in handling live retinas due to its gelatinous consistency and optical transparency. Although the protocol has been carefully optimized to reduce risks at this stage, damage to the inner retina, especially the ILM, remains possible. While this type of damage can be difficult to detect macroscopically, the close association of retinal astrocytes with the membrane means that major damage to the ILM is matched by corresponding damage to or ablation of astrocytes, as revealed with staining for the astrocyte marker GFAP (Figure 4K). Fortunately, it is not essential to remove the vitreous entirely for explants, although an excess can anchor the peripheral retina in a 'folded over' state, leading to metabolic damage from obstructed diffusion in vitro (see 'Metabolic injury' below). In order to visualize the vitreous and gain greater familiarity with its physical properties, one can use small volumes of a vital dye, such as 0.4% trypan blue31. However, although Food and Drug Administration (FDA)-approved formulations of trypan blue are widely used in ophthalmic surgery, the dye may impact downstream applications - particularly fluorometric and colorimetric assays - and we recommend restricting its use to training contexts.

Separation of the outer retina from the eye cup also poses a significant risk of piercing or tearing the retina, which can drive extensive RGC loss throughout neighboring regions and elevate microglia activation and astrocyte reactivity, and affected samples should be excluded from experiments. While this form of damage is often apparent macroscopically, more subtle injuries, such as penetrating incidents that did not fully pierce the retina but nonetheless drove poor experimental outcomes, were also encountered. Although such incidents may not be immediately apparent, a key characteristic of astrocytes in the retina and elsewhere is their response to mechanical injury, which includes upregulation of the intermediate filament GFAP. As such, staining with GFAP can reveal otherwise subtle mechanical injury (Figure 4J).

Finally, although relieving cuts are critical to allow the retina to lie flat in shallow media and ensure even diffusion of oxygen and nutrients, they are by nature a form of tissue damage, as evidenced by increases in GFAP expression and a decline in RGC density along the edge of the cut. Excluding these boundary regions from analysis when appropriate can minimize their effect on experimental readouts. However, care should also be taken when making cuts to ensure they produce relatively evenly-sized 'lobes', as we have found that particularly narrow lobes are characterized by attenuated central regions of consistently low glial activation/reactivity and relatively high RGC survival and, as such, may not be suitable for comparative analysis.

Metabolic injury
Because isolating and explanting the retina is itself a form of severe mechanical and metabolic injury, it can be challenging to identify and troubleshoot additional metabolic harms. An exception to this is curling or folding of the retina in culture, as this can be macroscopically apparent and produce localized zones of exaggerated damage, likely due to reduced diffusion of nutrients and oxygen. Typically, this folding is linked to excess vitreous near the edge of the retina, which can pull the tissue laterally or towards the optic nerve head, and often occurs when residual ciliary body is present at the retinal periphery. Therefore, despite the inherent risks of vitreal ablation (and the potentially salutary effect of retaining hyalocytes when leaving some vitreous) the amount of vitreous does need to be reduced to ensure that the retina lies relatively flat in vitro. We found that if the vitreous trapped in the angle between the ciliary body and retina is removed successfully (step 2.7 of the protocol), the retina can be mounted and cultured even if the ciliary body remains; in most cases, delicate brushing (or exceptionally careful use of forceps) can remove enough of this vitreous to preserve the sample.

Due to the challenge of identifying the origin of more diffuse metabolic injuries, we strongly recommend proactively minimizing risk by relying on fresh media. Supplemented explant media should be kept for no more than 1 week, and opened neurobasal media should be replaced after approximately 1 month. Eyes that have experienced extended postmortem intervals - 1 h or more at 4 °C, 30 min or more at RT - should be avoided for experiments that require culture, although those kept at 4 °C can be used for practice or shams over somewhat longer durations.

Figure 2: Changes in the organization and distribution of retinal glia after 3 days in vitro (DIV). (A) Iba1+ microglia in the RGC and nerve fiber layers form non-overlapping domains in the naïve mouse retina. (B) Retinal astrocytes express GFAP and are closely associated with the superficial vasculature, whereas the more abundant Müller cells are typically GFAP- in the naïve retina (sporadic Müller cell GFAP can be seen in the lower left corner of B and C). (C) Color composite of Iba1 (green) and GFAP (red) staining from the naïve retina in A and B. (D) After 3 DIV, microglial distribution in the explanted retina becomes irregular, suggesting migration, and microglia transition from a ramified morphology to a more amoeboid phenotype, characteristic of activation. (E) Retinal astrocytes show signs of reactivity (hypertrophy, diminished orientation to retinal vasculature) but no migration after 3 DIV, while Müller cells show greater GFAP expression, visible in the lower corner of E and F. (F) Color composite of Iba1 (green) and GFAP (red) staining from the 3 DIV retina in D and E. Scale bars = 100 µm. Please click here to view a larger version of this figure.

Figure 3: Changes to RGC density and glial morphology after 1 and 3 days 3 days in vitro (DIV). (A-C) Brn3a+ RGC density in the naïve mouse retina after sham explant (A), with ramified Iba1+ microglia (B) and vascularly associated GFAP+ astrocytes (C). (D) Color composite of Brn3a, Iba1, and GFAP from A-C. (E-H) After 1 DIV, Brn3a+ retinal ganglion cells decrease in density relative to naïve sham retinas, while the retracted processes of Iba1+ microglia (F) indicate ongoing activation; by comparison, changes in GFAP+ retinal astrocytes (G) are less pronounced, consisting primarily of modest hypertrophy. (H) Color composite of Brn3a, Iba1, and GFAP from E-G. (I-L) After three days in vitro, Brn3a+ retinal ganglion cells further decline in density (I), while Iba1+ microglia (J) continue transitioning to an activated, amoeboid morphology, and GFAP+ retinal astrocytes (K) show signs of further hypertrophy and reactivity, but lag changes seen in RGCs and microglia. (L) Color composite of Brn3a, Iba1, and GFAP from I-K. Scale bars = 100 µm for A - L. (M) Quantification of Brn3a+ RGC density in sham and 1-day in vitro retinas. Ganglion cell density was measured across 9 regions of interest per animal (n = 4 for each condition); counts from individual animals are displayed as filled circles, and error bars represent standard deviation. Please click here to view a larger version of this figure.

Figure 4: Changes in microglia and hyalocytes after 3 days in vitro (DIV) and use of GFAP staining to detect localized tissue damage in isolated retinas. (A-D) Iba1 strongly labels microglia in the naïve retina (A), while faintly labeling hyalocytes (A', inset), an additional myeloid cell type found at the vitreoretinal interface that expresses cell surface marker CD206 (B); naïve microglia express the homeostatic microglial marker TMEM119 (C). (D) Color composite of A-C highlighting colocalization of Iba1 (green) and TMEM119 (red) in microglia in naïve retina, and expression of CD206 (magenta) in an amoeboid hyalocyte (arrowhead). (E-H) After 3 days in vitro (DIV), Iba1+ retinal microglia withdraw their ramified processes and adopt increasingly amoeboid morphologies (E). CD206 expression remains confined to hyalocytes (F) (arrowheads), TMEM119 is no longer detectable in microglia (G). (H) Color composite of E-G, with Iba1+ (green) and Tmem119- (red) microglia in 3 DIV retina, and expression of CD206 (magenta) in amoeboid hyalocytes (arrowheads). Scale bars = 100 µm for A-H. (I-K) The response of retinal astrocytes to mechanical injury makes the astrocytic marker GFAP an excellent indicator of isolation-induced damage. (I) GFAP+ astrocytes, after 1 day in vitro, undergo modest hypertrophy but indicate no signs of physical injury. (J) GFAP expression in astrocytes and Müller cells reveals a site of isolation-associated injury in the retina, likely due to partial penetration of the retina by a metal instrument. Arrows denote the periphery of the injury area. (K) Severe disruption and loss of astrocytes (arrows) resulting from partial ablation of the inner limiting membrane at the retinal surface. Please click here to view a larger version of this figure.

Organotypic culture³² approaches such as retinal explants harness the experimental flexibility and rapid turnaround time of cell culture while preserving much of the in situ context of in vivo studies, making them a potent avenue for investigating complex interactions between cell types. We present this protocol as an entry point for researchers who may have substantial experience with the eye and be comfortable manipulating a fixed retina but not a fresh one; therefore, we have focused on reducing technical challenges and variation associated with isolation and handling, which our experience revealed as a major barrier to reliable generation of consistent samples. However, the eye is a complex organ composed of structures and tissues that vary nearly as much in material properties as they do in function, and the compact nature of the mouse eye further compounds the challenge of extracting the retina without damage. Therefore, a degree of flexibility is required to avoid excess structural or metabolic damage to the retina during isolation and reliably obtain high-quality tissue. Advice to recognize and minimize these complications is presented in the associated areas of the Results section ('Recognizing and troubleshooting mechanical injury' and 'Metabolic injury').

Although we and others have described the use of explants largely in the context of glaucoma - reflecting both our group's research focus and glaucoma's outsized role in driving vision loss - given reports of glial involvement in diabetic retinopathy^33,34, we anticipate this approach may be suitable for studying and manipulating these cells with regards to relevant aspects of this disease as well, which is a major cause of blindness in working-age adults. Additionally, due to the intrinsic characteristics of isolating the retina for explant, we anticipate that it would likewise serve as a valuable model of acute retinal trauma and ischemia and may provide insight into the role of retinal glia under these conditions. In characterizing this protocol, we have focused on time points of 1 and 3 days in vitro, partly because this is a sufficient duration to assess the success of the isolation and partly because we consider this initial window optimal for studying early changes in glial behavior. Beyond this time frame, neuronal loss driven by axotomy and the absence of retinal blood flow, along with photoreceptor death due to loss of RPE substrate, are expected to increasingly dominate outcomes. However, while the inherent challenges of keeping adult neural tissue alive ex vivo are substantial, others have reported maintaining explanted retinal tissue for longer durations^26,30,35under similar conditions, and it is informative to compare it with previously documented approaches for those who would like to experiment with keeping explanted retinas alive and potentially functioning for extended periods.

While a range of methodologies exist for culturing retinal tissue, we focus on those for isolating adult mouse or rat retinas with limited mechanical or chemical disruption. Perhaps the most influential of these, published in 2008, replaced serum supplementation of culture media with a widely used (including by this protocol) chemically defined media to maintain retinal tissue from adult rats for as long as 14 days²⁶. However, while the authors reported macroscopic patterns of tissue degradation similar to our observations at the cellular level - beginning at the periphery while central regions were initially spared - the use of transverse sections for histology makes direct comparison of RGC survival and changes to astrocytes difficult, and microglia remained uncharacterized. Later studies provide additional detail, including a 2019 study reporting functional characterization of explanted mouse retinas via electrophysiology³⁰. Although also following explanted retinas for up to 14 days, the authors documented a more rapid functional decline, with light-evoked photoreceptor response disappearing after the 2-day time point. Despite this loss, RGCs maintained attenuated responses to direct light stimulation through day 7, a phenomenon partially attributed to intrinsically photosensitive RGCs, which are both light-sensitive and relatively resistant to axotomy-induced degeneration³⁶. RGC loss at day 2 was alternatively reported as 37% or 21% (depending on methodology) compared with our assessment of a 33% decline at day 1, but counts of RGCs on naïve retinas yielded a density of 1,400 per mm² at baseline, far lower than both previous reports²⁴ (3,300 per mm²) and our own observations (3660 per mm²), all of which were derived from similarly aged C57BL/6J mice. Although differences reported in baseline densities complicate efforts to compare rates of RGC loss, directly comparable immunostaining of astrocytic GFAP showed results highly consistent with the observations at baseline and the 3-day time point, suggesting a similar degree of astrocyte activation (as with the initial 2008 study, microglia were not assessed).

A more recent study³⁵ potentially reconciles these divergent estimates of RGC density and loss by staining RGCs with both RBPMS and Brn3a (the former used in the 2019 study³⁰, the latter utilized in this study), as a statistically significant decline in RGC density was detected at an earlier time-point with Brn3a than RBPMS (day 3 versus day 5), and the loss measured by Brn3a was more substantial (44% versus 36%). It may be that expression of Brn3a - a transcription factor - is lost prior to RGC death in response to the acute injuries inherent to explant models. Regardless, RGC loss as measured by either marker appeared to plateau after this initial decline until at least day 7, and the authors suggest this period represents an optimal window for manipulations intended to improve RGC survival. In terms of glial changes, the authors reported evidence of astrocyte and Müller cell reactivity (GFAP upregulation) and signs of microglial activation (upregulation of Tnf-α, Il-1b, and Il-6) in their model, although these were only assessed at baseline and at day 7.

While methodological differences in assessment limit our ability to predict the outcome of keeping retinas isolated via this protocol for similarly extended periods, its general congruity with established approaches (regarding both initial isolation and culture conditions) suggests that relatively modest changes may facilitate longer-duration experiments. One notable difference between our approach and previous methods is the use of a filter as substrate throughout the culture period, which was avoided in this study due to the impact on downstream imaging. While we have not directly investigated the outcome of modifying the protocol to incorporate retention of the filter, given the vulnerability of photoreceptors, the provision of a lasting substrate during culture may be a valid approach for slowing degeneration of the explanted retina. Additionally, the most recent of the studies described above also found that supplementation of culture media with 200 ng/mL BDNF significantly improved RGC survival at the 5-day time point³⁵, suggesting an additional means for extending the temporal window of investigation. However, we would caution - given the potential confounding influence of additional media supplementation and the non-uniform pace and spatial distribution of degeneration reported in the explanted retina - that it is crucial to thoroughly consider which aspects of cell behavior or therapeutic intervention are of interest when determining experimental duration.

Although the primary aim of this study is to provide a detailed protocol for isolating retinal explants rather than a comprehensive guide to the experiments this approach enables, we would like to emphasize that while we consider immunofluorescence microscopy to be particularly useful for initial investigation and troubleshooting, explanted retinas are amenable to a range of assays (such as axonal regrowth²⁸ and multiplexed electrophysiological recording with multielectrode arrays³⁰) covering many aspects of cell behavior and tissue function. Likewise, while the intrinsic response of cells in the retina has been suitable for the current characterization, the accessibility of this tissue ex vivo makes it ideal for studying interventions that would be difficult or impractical to perform in live animals, particularly at scale. Some of the most striking applications of explanted retinas come from studies in which genetically targeted fluorescent protein expression has enabled real-time recording of glial behaviors such as migration²⁷ and phagocytosis³⁷, providing unprecedented insight into the workings of these understudied cells. We hope that our contributions here enable yet more researchers to probe the mysteries of non-neuronal cells in the retina.